By developing a large database of BAL fluid linked to specific microbiologic diagnoses, we plan to define protein expression signature response profiles that distinguish specific etiologies of lung infection and inflammation. These signature profiles will be based on mass spectrometry, two-dimensional gel electrophoresis and suspension array technologies. Because of the variability associated with individual host responses to infection due to differences in host immunity, sampling time effects, and external factors such as antibiotic or anti-inflammatory therapies, a large database will be required. The profiles of culture-negative BAL fluid will be of similar interest to assist in defining non-infectious etiologies of lung inflammation. A secondary objective is to perform proteomic analysis on serum collected from patients at the time of bronchoscopy. The goal is to link serum proteomic profiles to BAL proteomic profiles to determine whether a less invasive technique can predict infiltrate etiology with comparable sensitivity and specificity to BAL profiles. To complement the patient studies we have investigated protein biomarkers in blood and lavage from animal models of pneumonia. We have studied a rabbit model of invasive pulmonary aspergillosis (Proteomics 2010;10: 4270-4280) and a canine model of staphylococcal pneumonia (Am J Physiol Heart Circ Physiol 2007;293;H2487-500). Exploring these model systems will facilitate our identification of candidate biomarkers across species. We have recently developed new mass spectrometry-based protocols to detect bacterial peptides in bronchoalveolar lavage from patients with pneumonia. Identifying peptide biomarkers that are specific and unique for a pathogen offers the possibility of a method with higher sensitivity to detect bacteria in BAL. We have developed two methods (top down and bottom up approaches) that can be applied to clinical samples in order to rapidly identify Gram-negative pathogens. The former approach requires the removal of leukocytes, use of high performance liquid chromatography mass spectrometry (LC/MS) and deconvolution of the resultant ions to a database (Biotyper) that can be interrogated for the identification of the microorganism. The bottom up approach is based upon the creation of a theoretical tryptic core peptidome from genomic data and comparison with peptides generated from tryptic digests of bacteria that are analyzed by LC/MS-MS. The specificity of the peptides for a particular strain or species of bacteria is done by proteomic database analysis (Unipept, pBLAST) and then validated experimentally with lysed bacteria with labeled targeted peptides. We have recently used this approach to rapidly identify strain-specific peptide markers of Acinteobacter baumannii based on LC-MS/MS profiling of digested peptides (Clin Chem. 2016 Jun;62(6):866-75.). Clinical application of these methods is ongoing. We have a total of 589 clinical samples from 438 different subjects. Of the 438 subjects, 1 subject specified that their samples could not be shared with outside investigators. All subjects underwent the clinically indicated bronchoscopies and no samples were collected purely for research. The total number of pediatric subjects enrolled to date is 30. Approximately one half of the participants have a specific microbiologic diagnosis as a cause of their pulmonary infiltrates. Enrollment thus far in the cases of interest includes: Aspergillus species 58, P. jiroveci 63, Cytomegalovirus 61, Mycobacterial species 41, and Bacteria (gram negative or gram positive) 118. Approximately 50-60% of these infections occur with more than one microbial pathogen (i.e. Aspergillus species and cytomegalovirus). Infections with only a single pulmonary pathogen are somewhat less common than co-infection states. New samples of the bronchoalveolar lavage and blood are currently being analyzed.